There are two ways to arrive at a battery size. One is to look at the array and pick a pack that roughly matches a day's generation. The other is to look at what the house does between four and ten in the evening, and buy exactly that.
The second is nearly always cheaper, and it is the only one of the two that survives a winter.
Start with the meter, not the roof
Every smart meter in the country records half-hourly consumption, and your supplier will give you the data. That file is worth more than any rule of thumb.
- Export a full month. A winter month, if you have one — that is the load you are actually sizing for.
- Sum the half-hours between 16:00 and 22:00 for each day.
- Take the median of those daily totals, not the mean. One evening with the oven and the tumble dryer both running should not set your capital budget.
That median is your evening block. For most three- and four-bedroom houses without a heat pump it lands somewhere between 4 and 9 kWh. With a heat pump it can be double that.
Then compare against usable, not rated
A pack has two numbers and only one of them is yours.
- The 5 kWh residential pack is 5.12 kWh rated and 4.5 kWh usable.
- The 16 kWh pack is 16.07 kWh rated and 14.46 kWh usable.
The gap is depth of discharge — 90 % on both — and it exists so the cells are not routinely taken to the floor of their range. Sizing against the rated figure is how people end up with a system that runs out at nine o'clock every night in January.
Round-trip losses take a little more: 97 % on the smaller pack, 98 % on the larger. It is not much, but it means a 5 kWh evening block wants slightly more than 5 kWh of usable capacity behind it, not slightly less.
The arithmetic, once
Say the evening block comes out at 4.2 kWh. One 5 kWh pack covers it with margin. Say it comes out at 11 kWh — that is a 16 kWh pack, and a single 5 kWh unit would be a false economy you notice every day.
For the money, the shape is:
Annual saving ≈ evening block × 365 × (import rate − export rate)
Import and export rates are yours, not ours, and they move — put your own numbers in rather than trusting anyone's worked example, including this one. The subtraction matters more than either figure on its own: the battery is not earning you the import price, it is earning you the difference between buying that unit and selling it.
Then divide the installed cost by that saving. Anything you are shown that does not include the export rate in the subtraction is overstating the case.
What actually changes the answer
A tariff with a cheap overnight window. This is the single biggest lever, and it is free. If the pack can be filled overnight at a low rate, it covers the evening in December as well as it does in June, and the summer generation becomes a bonus rather than the whole business case.
A car. An 11 kW charger is 11 kW. It will not be run off the battery and should not be — it wants surplus generation and cheap hours, scheduled, and it belongs in the load study rather than in the battery sizing.
Expansion. Both residential packs parallel up to sixteen units on the same bus. Starting with one and adding another in two years costs a little more in total than buying two now, and considerably less than buying two you did not need. If the numbers are marginal, start small.
The part nobody quotes
Cycle life. Both packs are LiFePO4: 6,000 cycles on the 5 kWh unit, 4,000 – 6,000 on the 16 kWh. At roughly one cycle a day that is comfortably past a decade of daily use before capacity fade becomes the conversation.
A cheaper chemistry that gives you 2,000 cycles is not cheaper. It is the same purchase, made twice.
If you want this done against your own data rather than a worked example, send us a month of half-hourly export and we will do the sizing before anybody quotes you a number. Related: sizing storage against your own consumption.